Position Determines Perspective: Investigating Perspective Distortion for Image Forensics of Faces

Bo Peng; Wei Wang; Jing Dong; Tieniu Tan; Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2017, pp. 1-9

This paper points out a new telltale trace -- the characteristic of perspective distortion (CPD), for the image forensics of faces. The perspective distortion is determined by the position of image shooting, and it is often overlooked when creating a forgery, which results in the inconsistency between the claimed camera parameters and the CPD in the face image. To investigate this consistency problem, we cast it to the consistency between the claimed camera intrinsic parameters and the estimated ones from the CPD. Our parameter estimation approach is based on geometric observations that are related to CPD, like facial landmarks and contours. We analyze the estimation uncertainty caused by indeterminacy of observation to obtain a more reliable forensic decision. Experiments on synthetic datasets and real forgery examples demonstrate the effectiveness of the proposed method.